Geochemistry, Geophysics, Geosystems

Aeolian dust dynamics in central Asia during the Pleistocene: Driven by the long-term migration, seasonality, and permanency of the Asiatic polar front

Authors


Abstract

Loess-paleosol sequences preserve detailed archives of climate change, reflecting the dynamics of aeolian dust sedimentation and the paleodust content of the atmosphere. The detailed investigation of particle size distributions (PSDs) of windblown sediments is an increasingly used approach to assess the paleorecord of aeolian dust dynamics. The central Asian loess belt offers the potential to reconstruct Pleistocene atmospheric circulation patterns along an adjacent west-east transect within interior Eurasia through granulometric studies. In this study we present the aeolian dust record of the loess sequence at Remisowka (SE Kazakhstan), which reflects a detailed signal of glacial-interglacial climate and atmospheric dynamics in central Asia. On the basis of radiocarbon and amino acid geochronologic data, long-term semicontinuous trends in the aeolian dust record of the Last Glacial Cycle are observed and interpreted to reveal their paleoclimate signal. In consideration of the modern synoptical atmospheric circulation patterns and aeolian dust transport in central Asia, it is likely that the observed trends reflect the long-term migration, seasonal duration, and permanency of the polar front during the late Pleistocene. Previously published models, which focused on the reciprocal glacial-interglacial influence of the zonal Westerlies and the Asiatic high on the aeolian dust transport in central Asia, were overly simplified and should be modified to include the major influence of the Asiatic polar front. As the polar front activity is intimately connected with the development and position of the interhemispherically active, high-level planetary frontal zone (HPFZ), the presented data give insight to long-term aeolian dust dynamics and climate variability of interior Eurasia, which are linked with interhemispheric climates.

1. Introduction

Aeolian mineral dust plays an important role in the atmospheric system and has a significant influence on the radiation balance of the Earth. Loess deposits were formed by the more or less continuous accumulation and postdepositional modification of aeolian dust in continental midlatitudes during the Quaternary Period. The type of accumulated airborne dust, its origin, and the natural mechanisms that supply, deposit, and pedogenically alter the material are driven by dynamic processes at the land-atmosphere interface, and are dependent on general paleoclimatic conditions [Smalley, 1995; Assallay et al., 1998]. The latter information is recorded during the deposition of loess. Loess-paleosol sequences are therefore detailed archives of climate change, recording the dynamics of aeolian dust sedimentation and the paleodust content of the atmosphere. Studying the loess record may provide insight into past atmospheric circulation patterns and synoptic climate variations.

When reconstructing past atmospheric circulation on the basis of widespread aeolian dust deposits, such as loess, complex atmospheric flows and paleoclimatic conditions have to be considered as they play a crucial role in the entrainment and aeolian transport of mineral material by defining the general flow direction and the strength of near surface winds [Pye, 1995]. Humidity, temperature, surface roughness and seasonal flow patterns also have to be taken into account.

Central Asia is one of the most significant loess regions on Earth, with loess sequences more than 200 m thick found on the foothills and forelands of the high mountain chains [e.g., Frechen and Dodonov, 1998]. The distribution of central Asian loess deposits, located between the well-studied European loess sequences to the west and the extensive Chinese loess plateau region to the east, enables researchers to carry out interregional paleoclimatic investigations along a west-east transect across the entire Eurasian loess belt of the Northern Hemisphere (see Figure 1) [Penck, 1930; Mavlyanov et al., 1987; Dodonov and Baiguzina, 1995; Dodonov, 2002]. Dust sedimentation in central Asia took place under predominantly semiarid and arid climatic conditions during glacial periods, forming thick loess units that indicate long periods of dust accumulation [Penck, 1930; Dodonov, 2002]. During warmer, slightly more humid interglacial intervals of the Pleistocene, climatic conditions led to reduced silt deposition and enhanced soil formation of Kastanozems, Serosems and other steppe-like soils. However, the soil formation within one warm period is often not continuous. Phases of soil formation alternate with phases of loess sedimentation, forming intercalated loess layers that indicate climatic fluctuations during an interglacial period. Both the paleosols and the intercalated interglacial loess layers form characteristic pedocomplexes, such as those correlated with marine oxygen-isotope stages (OIS) 5 or 7, which are observed in numerous loess profiles throughout Eurasia. High sedimentation rates during glacial periods, alternating with low translocation and weathering rates during interglacials, as well as the absence of periglacial features, provide a pristine and detailed aeolian record of Pleistocene climate variability in Eurasia.

Figure 1.

Map of central and middle Asia (after Machalett et al. [2006] with permission from Elsevier) showing the major loess belt with deposits up to 200 m thick on the foothills and forelands of the high mountain chains and the location of the investigated loess-paleosol sequence nearby Almaty, SE Kazakhstan.

Detailed investigation of particle-size distributions (PSDs) of windblown sediments is an increasingly used approach in assessing loess records of aeolian dust dynamics [e.g., Prins and Vriend, 2007; Prins et al., 2007]. The grain size of the transported dust is usually associated with wind speeds and/or wind strengths. As a simplified ideal system, it can be assumed that relatively coarse grain-size fractions in loess (medium and coarse silt to fine sand) reflect dynamic environmental conditions typical for colder, glacial climates and local or regionally derived sediment, transported by prevailing strong winds. In contrast, smaller grain-size fractions (fine silt and smaller) are related to less dynamic, primarily warm climatic conditions and are typical of dust loading in the upper atmosphere representing transport over long, in certain cases interhemispheric, distances [cf. Kumar et al., 2003]. Ding et al. [2005] and Han et al. [2007] emphasized that the variations in dust grain-size and mass accumulation rate during interglacial-glacial periods lead not only back to changes in wind speed, but changes in the source-to-sink distance that must have also played a dominant role. Recently Prins et al. [2007] offered a new approach to get a better understanding of the variability of aeolian dust input by unmixing the grain-size records (end-member modeling) of loess samples taken from the Chinese Loess Plateau.

Often it has been hypothesized that grain-size variations in the loess record are the consequence of different, perhaps alternating or competing, atmospheric circulation patterns or airflow systems. Prominent examples are the grain size and magnetic susceptibility variations analyzed from the Chinese Loess Plateau [An et al., 1991; An, 2000; Liu and Ding, 1998]. These are interpreted as a direct proxy for the strength of the dust-bearing east Asian winter monsoon [Kukla, 1987; Derbyshire et al., 1995; Vandenberghe et al., 1997], alternating with the warmer, wetter periods of enhanced pedogenesis driven by the summer monsoon [An et al., 1991].

On the basis of variations in grain-size records of the last glacial cycle, several authors suggested that short-term climatic events originating in the North Atlantic, e.g., Dansgaard-Oeschger events, are recorded in loess sequences of Eurasia [Porter and An, 1995; Chen et al., 1997; Vandenberghe and Nugteren, 2001; Rousseau et al., 2002]. However, dating techniques available for assessing the ages of loess sequences and their derived paleoclimate proxies, e.g., luminescence dating, have mean standard deviations of about ±10%. Thus the resolution is too low to determine precise ages of brief duration events. Nevertheless, Ruth et al. [2007] found evidence for a strong link between North Atlantic and east Asian glacial climates by interpreting the Greenland ice core dust record and the grain-size record of the Chinese loess plateau, which corroborates the inference of an intrahemispheric link for the aeolian dust dynamics of the Northern Hemisphere.

Despite the high potential of the central Asian loess for granulometric investigations and reconstruction of past atmospheric circulation patterns, only a few loess studies in southern Tajikistan have been carried out, focusing on highly resolved grain-size dynamics [e.g., Ding et al., 2002]. Almost all presented concepts follow the scheme introduced by Dodonov and Baiguzina [1995], suggesting that during cold cycles the winds from the Asiatic high (also known as the Siberian-Mongolian High) dominated and transported mainly coarse dust, while during warm cycles mainly the Westerlies controlled dust sedimentation through the long-distance transport of finer dust to central Asia.

Regardless of successes of previous investigations, the climatic forcing factors driving aeolian dust dynamics in Eurasia are not fully understood, and the teleconnection of climatic events across the Northern Hemisphere during the Pleistocene is still a matter of debate. In this study we present the first highly resolved grain-size data from the late Pleistocene loess record in SE Kazakhstan. With support of radiocarbon and amino acid geochronology data the dynamics of the past aeolian dust transport are discussed against the background of modern synoptic climatic patterns of this region. Our aim is to show that hitherto discussed past atmospheric circulation models in central Asia were overly simplified, as they focused only on the glacial-interglacial reciprocity of the zonal Westerlies and the Asiatic high. We state, in consideration of the modern synoptic atmospheric circulation patterns and aeolian dust transport in Eurasia, that past and modern dust dynamics in central Asia are mainly triggered by changes in duration and permanency of the Asiatic polar front. We discuss the significant long-term climatic trends within the grain size record of the last glacial cycle and show their connection with the long-term seasonal shift of the Asiatic polar front.

2. Climatic Setting and Atmospheric Dust Transport in Central Asia

Central Asia is well defined both geographically and climatically and has a unique “self-contained” climate. Located in the inner part of the Eurasian continent, remote from the oceans and surrounded by high mountain ranges, the region is characterized by an extreme type of the temperate continental climate [Franz, 1973]. The low latitudinal position results in long sunshine duration and limited cloudiness. Mean summer solar radiation is high, resulting in high temperatures during the warm season, while in the winter season very cold air can invade central Asia because of the open exposure to the north and northwest (see Figure 1) [Belyaev, 1995; Lydolph, 1977; Weischet and Endlicher, 2000].

The configuration of the atmospheric circulation over the region is considerably affected by two natural climatic boundaries. The Caspian Sea generates interaction with the Caucasian Highlands, forming a distinct barrier in the west and precludes the inflow of cyclonic storms from a westerly or northwesterly direction. The high mountain system surrounding central Asia on its southern circumference inhibits the intrusion of air masses from the south, e.g., the Indian monsoon system. Moreover the mountain ranges exert a significant impact by splitting the zonal Westerlies into a southeastward flow over northern India and a northeastward flow over northern China. This perturbation in the troposphere generates complex patterns of divergence and convergence in the upper air and affects the position of the so-called “high-level planetary frontal zone” (HPFZ) [cf. Lydolph, 1977; Bugaev et al., 1957]. This upper tropospheric feature has a great influence on the development of the Asian sector of the polar front with its associated wave and cyclonic activities. The frequency of cyclonic storms is linked with the seasonal shift of the Polar front from north to south and vice versa, which distinctively forces the annual distribution of rainfall and the transition from the cold to the warm season in central Asia [Franz, 1973; Lydolph, 1977]. During winter the Asiatic polar front, the western branch also known as the Iranian front, is mainly located over the highlands of Iran, Afghanistan and the central Asian mountains and causes precipitation and storminess on the southwesterly and westerly slopes of the mountains (Figure 2c). From February onward the Polar front shifts to the north resulting in vigorous cyclonic storms affecting the deserts, steppes and the foothills of the mountains, bringing a striking spring maximum of precipitation from late March to June. In summer the polar front is rarely developed owing to the weak contrast of the continental air on both sides of the front. A second maximum of the frontal activity takes place in autumn when the polar front shifts back to the south [Franz, 1973; Belyaev, 1995].

Figure 2.

Percentage frequencies of surface airflow in central Asia after Lydolph [1977] (with permission from Elsevier) and position of the Asiatic polar front during winter after Franz [1973]. Central Asia experiences a large variety of surface airflow and pressure patterns in all seasons. (a) The summer flow aloft is more zonal, dominated by westerly, northwesterly, and northerly intrusions and thermal depressions. (b) In the winter months cyclonic storms are more frequent and are, together with the frontal wave activity at the Asiatic polar front, the main forces of aeolian dust transport and dynamics in the region. (c) During winter the Asiatic polar front is mainly located over the central Asian mountains and causes precipitation and storminess on the southwesterly and westerly slopes of the mountains.

In spite of the general zonal flow during the year, central Asia experiences a large variety of surface airflow and pressure patterns in all seasons. Bugaev et al. [1957] distinguished 12 types of surface airflow in central Asia, with variable percentage frequencies of occurrence in summer and winter (Figures 2a and 2b). The summer flow aloft is more zonal, dominated by westerly, northwesterly and northerly intrusions and thermal depressions, while in the winter half-year cyclonic storms are more frequent.

Three major types of cyclones, the South Caspian, the Murgab and the Upper Amu-Darya, penetrate into central Asia and are, through their interaction with the seasonal shift of the Asiatic polar front, the main forces of aeolian dust transport and dynamics in the region, next to the frontal wave activity and cyclonic circulation along the polar front itself [Lydolph, 1977; Weischet and Endlicher, 2000]. When a lack of precipitation and low absolute humidity prevails, these cyclonic storms often result in hazardous dust storms that mainly affect the forelands of the mountains ranges. During summer the dust transport is mainly caused either by thermal depressions that lead to turbulence along the orographic obstacle of the high mountains or cold outbreaks from the north. Even though there is prevalent dust transport in the higher atmosphere, which determines the hazy, whitish-hue sky in central Asia, the outflow, transport and deposition of aeolian dust is mainly a regional or local phenomenon [Lydolph, 1977; Machalett et al., 2006]. It was already stated by Penck [1930] that “different causes – glaciers, rivers, and wind, nival and arid conditions” as well as vegetation “must work together” to produce silty mineral material, bring it in suspension as airborne dust and lodge it on certain places as loess. The dust storm in Turkmenistan on 20 January 2007, caused by the Murgab cyclone that came out of Iran, exemplifies the concurrence of mountain glaciers, deserts and steppes, river valleys and alluvial fans, orographic obstacles, and the small temperate belt along the foothills between the snow and arid line for the aeolian dust transport and sedimentation in central Asia (Figure 3 with further explanations). The occurrence of this suite of geomorphologic systems reveals central Asia as an archetypical landscape for the study of aeolian dust dynamics.

Figure 3.

Dust storm in Turkmenistan on (a) 26 December 1937 (after Lydolph [1977] with permission from Elsevier) and on (b) 20 January 2007 (MODIS scene and Blue Marble background taken from NASA WORLDWIND 1.4) caused by the Murgab cyclone that occurs principally in the winter months. When a lack of precipitation and low absolute humidity prevail, cyclonic storms such as the Murgab cyclone, driven by the winter temperature and pressure gradient between warmer moisture tropical air and colder drier polar air along the polar front, often result in hazardous dust storms that mainly affect the forelands of the mountains ranges.

3. Study Area

In order to explore the loess sedimentary record of past atmospheric dust dynamics in the region, we investigated the loess-paleosol sequence at Remisowka, southeastern Kazakhstan, nearby the former capital Almaty (Alma Ata), located in the foothill zone of the northern Tien Shan (Figure 4). The loess belt in the proximity of Almaty is one of the most promising widespread terrestrial climate and environmental archives of the Pleistocene in central Asia, in addition to the loess of southern Tajikistan and the loess in the region of Tashkent and Samarkand in Uzbekistan [Mavlyanov et al., 1987; Pécsi and Richter, 1996; Smalley et al., 2006]. Despite some early loess studies in this region [e.g., Kolotilin, 1953; Lomonowitsch, 1953, 1955] and first studies of the proximal loess province in North Xinjiang by Qizhong and Honghan [1988], more recent studies of the loess record of SE Kazakhstan have not yet been published. Machalett et al. [2006] provided a first description of the loess deposits at the Remisowka site, where they applied luminescence dating and underlined the key role of the Kazakh loess as a link between the central Asian and Chinese loess provinces.

Figure 4.

(a) Geological cross section of SE Kazakhstan with the location of the Remisowka loess sequence and (b) picture showing the upper part of the studied sequence. Note the suite of geomorphologic systems that all contribute to keep the central Asian dust cycle running [Smalley et al., 2006]. Silty and fine-grained material produced in the mountains is carried out into the alluvial fans, the desert, and steppes by mountain rivers. From there an outblow can occur and lodge the airborne material on the small temperate belt along the perimontane foothills. The sequence under study, located at an altitude of about 1070 m above sea level, is part of the Kamenskoe loess hills, which are bounded by the alluvial fans of the Bolshaja Almatinka and Malenkaja Almatinka rivers. The loess thickness at the Remisowka section reaches 80 m. This investigation focuses on the upper part (28.5 m) of the 50 m exposed loess sequence.

Southeastern Kazakhstan experiences the typical central Asian climate. Stable, hot and dry summers alternate with cold winters. Both the time of maximum precipitation and the major occurrence of regional dust storms in spring and autumn are connected with the seasonal shift of the Asiatic polar front [Bugaev, 1946]. Although glacial and interglacial atmospheric dust loading is very different and the source-to-sink distance must have played a critical role for changes in the regional or local aeolian dust activities on glacial-interglacial timescales, Pleistocene as well as modern mineral dust was mainly derived from river valleys and alluvial fans at the mountain fringe and from the surrounding steppes and deserts, e.g., the Muyunkum desert or Taukum desert [Lomonowitsch, 1955]. The deposition of the airborne dust occurred mainly on the perimontane foothill zone, where loess thickness can reach 80–100 m [Walter, 1974].

The sequence under study (Figure 4), located at an altitude of about 1070 m above sea level, is part of the Kamenskoe loess hills, which are bounded by the alluvial fans of the Bolshaja Almatinka and Malenkaja Almatinka rivers. The loess thickness at the Remisowka section reaches 80 m. This investigation focuses on the upper part (28.5 m) of the 50 m exposed loess sequence (see Figure 4 and Figure 5). Below the modern soil, three loess layers could be distinguished, intercalated by a weak pedocomplex between 6.80 and 11.25 m below the surface and a compound pedocomplex between 17.90 and 23.70 m depth. The latter pedocomplex consists of a strongly developed, doubled paleosol horizon and two superposed weakly developed paleosols. Our present study ends within the third loess layer in the profile, as so far we have only been able to develop a reliable chronological framework for the upper part of the total sequence. Machalett et al. [2006] provides a detailed description of the geology and stratigraphy of the site.

Figure 5.

Chronostratigraphic scheme of the studied sequence based on the results of radiocarbon dating and amino acid geochronology. Radiocarbon ages (Table 1) shown are uncalibrated 14C years B.P. D/L glutamic acid data are mean and one standard deviation total acid hydrolysate values measured in Pseudonapaeus retrodens shells. Summary D/L data are shown in Table 2. Below the modern soil, three loess layers could be distinguished, intercalated by a weak pedocomplex between 6.80 and 11.25 m below the surface. A compound pedocomplex exists between 17.90 and 23.70 m depth. Considering the bracketing 14C ages, the first weak pedocomplex is interpreted to correlate with the interstadial complex correlated with OIS 3, overlying a loess layer that very likely correlates to OIS 4, which is in conjunction with the results of amino acid geochronology that show a small but distinct increase in the D/L ratios from the two uppermost samples to the samples taken from the loess layer below the weak pedocomplex. The significant increase in D/L ratios measured in samples from loess above and below the strong, lower pedocomplex reveals a clear temporal distinction between those loess layers, indicating the position of the last interglacial interval in the Remisowka sequence. We interpret the lowermost doubled soil horizon to correlate with OIS substage 5e, and the overlying weak paleosols represent late OIS 5 substages 5c and 5a.

4. Material and Methods

A continuous 3–4 m deep trench (see Figure 4) was prepared for field sampling in 2004 and 2006, which allowed for collection of pristine sample material that had not been influenced by recent bioturbation or surface weathering. After a detailed cleaning and description of the section, bulk samples for particle size analysis were collected in 10 cm and 2 cm intervals. For radiocarbon dating and amino acid geochronology bulk samples were collected from narrow intervals (15–20 cm) within the loess layers, bracketing the paleosols, and were washed through a 1 mm sieve to recover fossil gastropod shells.

4.1. Grain Size Analyses by Laser Diffractometry

Laser diffractometry has become accepted to measure particle-size distributions (PSDs) as it offers important advantages compared with traditional sieve-pipette methods or image particle analysis [cf. Stuut, 2001; Stuut et al., 2002]. In this study, particle size distributions of grain size samples were measured using a Beckman-Coulter LS 13320 PIDS laser diffraction particle size analyzer. An auto-prep station was used to measure each sample (5 aliquots per sample) under equal conditions to prevent random operator errors. Fifty test samples were measured under different conditions in order to develop an optimized sample dispersion and standard operating procedure, taking into account the influence of organic matter, carbonate content and particularly mechanical (e.g., ultrasonic) pretreatment. Altogether more than 1500 test measurements were made in the process of developing a standard sample protocol. Similar to Beuselinck et al. [1998], we observe that the effect of the removal of organic matter is negligible. Furthermore, the content of organic material is very low, ranging from 0.6% to 1.5% within loess and paleosols, respectively. On the basis of the test results, we elected not to remove organic matter from the stratigraphic samples. We also rejected the removal of carbonates, because they are a primary component of the sediment formation cycle of loess in SE Kazakhstan and cannot be separated from secondary carbonates.

In contrast to several other granulometric studies we chose not to apply an intensive ultrasonic treatment to disaggregate particles, and instead applied an effective, but gentle, procedure of dispersion by spiking each sample with 1% ammonium hydroxide and treating them for at least 12 hours in overhead tube rotators. Ultrasonic disaggregation has significant drawbacks because it can lead to reaggregation, cleave weak mineral grains, and produces air bubbles, causing ghost signals during the laser diffraction analyses. Our test results suggest that grain size measurements using traditional intensive ultrasonic treatment, as reported in previous studies, should be reevaluated.

4.2. Amino Acid Geochronology

Paleoclimatic investigations of loess-paleosol sequences depend on the application of numerical dating techniques, such as radiocarbon and luminescence methods, in order to develop reliable time series for the proxies being studied. Commonly, the utility of luminescence and radiocarbon dating is limited by their applicable dating range or, in case of some central Asian loess sites, results show a significant age underestimation for samples taken from the last glacial cycle [Zhou et al., 1995; Machalett et al., 2006]. Relative dating approaches, such as amino acid geochronology, offer an independent assessment of numerical age estimates, when results are at or near their methodological limits, and assist in the chronostratigraphic evaluation of loess units beyond the range of useful numerical dating methods. Amino acid geochronology has been successfully applied to fossil gastropod shells from calcareous loess deposits from loess-paleosol sequences in North America and Europe, [Oches and McCoy, 1995, 2001; Oches et al., 1996, 2000], and China [McCoy et al., 1988; Oches and McCoy, 2001]. In this study, we present the first aminostratigraphic results from a central Asian loess site.

Amino acid geochronology measures the extent of racemization of amino acids within the carbonate shells of fossil molluscs. Amino acids, which exist in the levorotatory (L-form) optical configuration in living organisms, play an essential role in the biomineralization process of shell formation. Following protein synthesis, amino acids encased within the shell crystal structure, undergo reversible stereochemical inversion, or racemization, to their dextrorotatory (D-form) enantiomers. The rate at which the L- to D-inversion occurs is primarily a function of environmental temperature, and the extent of racemization, measured by the D/L ratio, is dependent on the amount of time that has passed. Taxonomic factors may significantly influence racemization; therefore sample sets of a single genus are typically analyzed in aminostratigraphic studies. Thus, within a monogeneric suite of fossil shells that have experienced similar postdepositional temperature histories, the D/L ratio measured in individual amino acids can be a useful measure of age. Terrestrial gastropod shells are often abundant and well preserved in loess deposits, making them an ideal system for the application of this method. A detailed review of the principles and applications of amino acid geochronology is available from Wehmiller and Miller [2000].

In this study we present results of D/L glutamic acid ratios measured on fossil shells of the terrestrial gastropod Pseudonapaeus retrodens [Martens, 1879]. We have measured additional amino acids, including D/L-aspartic acid, phenylalanine, valine, and alloIsoleucine/Isoleucine, although those results will be presented in detail in a later publication. Pseudonapaeus is widespread in central Asian loess sequences [Meng, 2004] and was present in most sampled loess layers at Remisowka. Taxonomic identification of the sieved fossil snail shells were determined by the comparison with originally described type examples of the Malacozoological Collection at the Museum für Naturkunde of the Humboldt-University of Berlin. The samples were prepared and measured at the Amino Acid Geochronology Laboratory at the University of South Florida using reverse-phase liquid chromatography, following the method described by Kaufman and Manley [1998].

5. Chronostratigraphic Scheme

The luminescence dating results for the Remisowka sequence, reported by Machalett et al. [2006] revealed methodological problems in IRSL dating of the polymineral fine-grain fraction applying the multiple aliquot additive dose (MAAD) protocol. It is apparent that anomalous fading of the luminescence signal within feldspar minerals resulted in age underestimation for samples from throughout the profile. This appears to be a common problem when using feldspar minerals for luminescence dating [cf. Lai and Brückner, 2008]. Similar results were reported by Zhou et al. [1995] for the sequence Orkutsay in Uzbekistan. Given age underestimation problems, we interpret the results to indicate that the first weak pedocomplex (6.8–11.25 m) represents an interstadial period that correlates with OIS 3.

To verify the chronostratigraphic position of the OIS 3 complex, radiocarbon dating on fossil gastropod shells taken from the uppermost loess layers was carried out at the Leibniz Institute for Applied Geosciences in Hannover, Germany. All age estimates were calibrated using the calibration program as described by Fairbanks et al. [2005]. Results are shown in Table 1. The first sample, taken from the uppermost loess layer below the modern soil, yielded an age of 24655 ± 764 cal. B.P. (Hv-25415). A second sample within this loess layer, collected close to the top of the first pedocomplex, dated to 28391 ± 1402 cal. B.P. (Hv-25416). On the basis of these age estimates, the uppermost loess layer can be correlated with OIS 2 and is characterized by high sedimentation rates of aeolian dust (see Figure 5). Two samples obtained from the bottom of the weak pedocomplex produced ages that are close to or already above the upper reliable radiocarbon dating limit. Sample Hv-25418 yielded an age of 37452 ± 2661 cal. B.P., while sample Hv-25417 is beyond the dating limit (>36,000 B.P.). The age underestimation for that sample is verified by sample Hv-25419 (>36,000 B.P.) taken from the loess below. Considering these bracketing ages, we interpret the first weak pedocomplex to correlate with the interstadial complex correlated with OIS 3, overlying a loess layer that very likely correlates to OIS 4.

Table 1. Results of Radiocarbon Datinga
Sample Depth (m)Laboratory NumberMaterialδ13C (‰)14C age (Years B.P. ± 1σ)Calibrated Age (Mean) (in Years Cal. B.P. ± 1σ)
2.60–2.80Hv-25415snail shells−4.220635 ± 59024655 ± 764
5.80–6.00Hv-25416snail shells−5.523670 ± 117028391 ± 1402
10.85–11.05Hv-25417snail shells−5.5>36000-
10.85–11.05Hv-25418snail shells−5.732100 ± 269037452 ± 2661
17.55–17.75Hv-25419snail shells−6.1>36000-

The results of amino acid geochronology (see Table 2 and Figure 5) provide an independent correlated age model for the profile. The D/L ratios of glutamic acid show only a slight increase in the upper loess layer from 0.145 ± 0.007 in sample FAL 1898 to 0.168 ± 0.014 in sample FAL 1900. From the two uppermost samples to the samples taken from the loess layer below the weak pedocomplex, a small but distinct increase in the D/L ratios occurs. D/L glutamic acid ratios of the underlying samples, FAL 1927, FAL 1938 and FAL 1932, range on a plateau from 0.193 ± 0.004 to 0.199 ± 0.010. This shift in D/L values from sample FAL 1900 to sample FAL 1927 suggests that the upper weak pedocomplex represents a brief weathering interval during which temperatures remained relatively cold, consistent with an interstadial period. In conjunction with the radiocarbon age estimates, we conclude that the upper pedocomplex correlates with OIS 3.

Table 2. Results of Amino Acid Geochronologya
Sample Depth (m)FAL Laboratory NumberD/L - Glutamic Acid (Mean ± Std. Dev.)Number of Subsamples
  • a

    Summary total hydrolysate amino acid racemization data for D/L-glutamic acid measured in Pseudonaepus retrodens shells from loess at Remisowka, SE Kazakhstan.

2.60–2.80FAL 18980.145 ± 0.0074
5.80–6.00FAL 19000.168 ± 0.0146
10.15–10.35FAL 19270.193 ± 0.0043
14.60–14.80FAL 19380.199 ± 0.0104
17.55–17.75FAL 19320.195 ± 0.0095
21.50–21.65FAL 19100.229 ± 0.0163
24.60–24.80FAL 19430.2851
28.05–28.25FAL 19450.327 ± 0.0233
28.35–28.55FAL 19090.335 ± 0.0102

Comparing D/L glutamic acid ratios measured in samples from loess above and below the strong, lower pedocomplex a clear temporal distinction between those loess layers is revealed, allowing us to confidently determine the position of the last interglacial period in the Remisowka sequence. The first striking jump in D/L glutamic acid ratios is registered at the transition from the loess above and below the strong pedocomplex. A greater than 60% increase of D/L ratios is observed from sample FAL 1932 (0.195 ± 0.009) to FAL 1945 (0.327 ± 0.023) and FAL 1909 (0.335 ± 0.010). This significant increase in D/L ratios argues for a long period of increased temperature, with distinctly lower rates of aeolian dust sedimentation. As there are no observed erosional discontinuities in the sequence, it is suggested that the whole pedocomplex (17.90 to 23.70 m depth) formed during the last interglacial period (sensu lato). Thus, it becomes very likely that the lowermost doubled soil horizon (22.25 to 23.70 m depth) correlates to OIS substage 5e. On the basis of amino acid racemization data, we correlate the loess layer below that pedocomplex with OIS 6, which is in disagreement with our previous preliminary geological age estimates [Machalett et al., 2006]. The overlying weak paleosols appears to represent late OIS 5 interstadials (substages 5c and 5a) supported by the small increase in D/L ratios in sample FAL 1932 (0.195 ± 0.009) and sample FAL 1910 (0.229 ± 0.016), compared with results from the loess above (see Figure 5).

On the basis of the results of radiocarbon dating and the aminostratigraphic scheme, the upper 28.50 m of the Remisowka loess sequence represents a well-resolved paleoenvironmental record of the late Pleistocene, suitable for a detailed reconstruction of the climate of the last interglacial/glacial cycle.

6. Granulometric Features and Sedimentological Parameters

The PSD of the SE Kazakhstan loess depicts a typical loess sediment, with a characteristic polymodal distribution and a strong dominance of the middle and coarse silt fraction; a representative PSD is given in Figure 6a. It shows two distinctive maxima in the silt fraction, in the finer and coarser silt sizes (peaks B and C), that are separated by a noticeable gap centered on 29 μm. The grain-size curve has a clear shoulder (peak A) in the area of the clay fraction and falls abruptly through the coarser silt and sand fractions. In some samples a fourth small peak D is visible within the fine sand fraction around 63 μm and 120 μm. The content of fine sand seldom rises to more than 6–8%. Clay content (<5.61 μm) is generally between about 19 and 27%, although within the paleosols it can reach more than 40%. Mean values of all measured samples span a range from 15.5 μm to 32.5 μm. The mean values for the loess are coarser than those for the paleosols.

Figure 6.

Typical particle size distributions (PSDs) of the SE Kazakhstan loess, taken from different stratigraphic layers of Remisowka and applied grain size parameters. PSDs of grain size samples were measured by a Beckman-Coulter LS 13320 PIDS laser diffraction particle size analyzer. (a) A special characteristic of the examined sample set is the occurrence of the bimodal distribution (twin peaks) within the silt fraction. While both maxima (peak B and peak C) vary significantly in height and area within the PSDs from the profile, the position of the gap is consistent at around 28–30 μm. In order to depict the variations within the bimodal silt distribution we calculated the Twin Peak ratio (11.8–27.4 μm/30.1–63.4 μm), establishing a quantitative comparison between silt peaks B and C. (b) In order to compare with other studies, we calculate the U ratio (15.6–63.4 μm/5.61–15.6 μm [Vandenberghe et al., 1997]). (c) The Uf−m ratio, defined as the ratio between the fine and medium silt fraction, is a representation of the morphology of the PSDs in the SE Kazakhstan loess.

The latter pattern is underlined when applying grain size parameters or ratios that focus on the silt and sand-sized fraction within the PSDs, which is a common method to clarify the sedimentological features of aeolian sediments and to eliminate the influence of postdepositional processes [Liu, 1988; An and Porter, 1997; Vandenberghe et al., 1997; Antoine et al., 2001]. Giving the possibility of comparison with other grain size studies we applied the U ratio (15.6–63.4 μm/5.61–15.6 μm), which expresses the ratio between the medium-coarse and the fine fraction of the silt-sized range and calculated the volume percent of grain size fractions > 63 μm (Figure 6b) [Vandenberghe et al., 1997]. Furthermore in consideration of the individual morphology of the PSDs in the SE Kazakhstan loess we applied the Uf−m ratio, which is a ratio between the fine and the middle silt fraction (Figure 6c).

A special characteristic of the examined sample set is the occurrence of the bimodal distribution (twin peaks) within the silt fraction. While both maxima (Figure 6a; peak B and peak C) vary significantly in height and area within the PSDs from the profile, the position of the trough is consistent at around 28–30 μm. The variations of the twin peaks are not random and reflect a signal that very clearly correlates with the stratigraphic position and the paleoclimatic features through the whole loess sequence from Remisowka. The coarser silt peak C is pronounced within loess layers, while the finer silt peak B is stronger in stratigraphic layers that correspond with temperate climate conditions, e.g., pedocomplexes and adjoining units (see Figure 6). It is striking that both peaks appear in almost the same height and area within transition horizons from paleosols to loess units (and contrariwise) as well as in intercalated loess layers within the pedocomplexes. In order to depict the variations within the bimodal silt distribution we calculated the Twin Peak ratio (11.8–27.4 μm/30.1–63.4 μm), establishing a quantitative comparison between silt peaks B and C (Figure 6a). A smaller ratio stands for a predominant coarser silt deposition and a higher ratio for prevailing finer silt deposition. It should be noted that previous investigations probably could not see this phenomenon in the silt range of aeolian sediments, as a high resolution of the used particle size analyzer and a careful sample preparation are necessary to discriminate and resolve both peaks. We could exclude all possibility of doubt that the bimodal distribution could be a measurement artifact, by remeasurements on different machines of the applied Beckman Coulter particle size analyzer, investigations on an CILAS particle size analyzer (K. O'Hara-Dhand et al., Particle size analysis of loess from Ruma brickyard and Titel Plateau, Vojvodina, Serbia, submitted to Sedimentary Geology, 2008) and the measurements of several other continues sample sets from Europe, central Asia and the Chinese Loess Plateau, which show a similar phenomenon and will be presented in a later publication. Even though a potential influence of secondary carbonates on the analyzed PSDs cannot be excluded, we found that the removal of carbonates result in a decline in reproducibility and deletion of some grain-size fractions, but do not affect the setting of the bimodal silt distribution or the variations between peak B and C. Additional studies suggested that the twin peak phenomenon seems to be intimately linked to the crystal structure and complex deformation mechanisms of the alpha quartz minerals, a main constituent for loess particles [Machalett et al., 2008; O'Hara-Dhand et al., submitted manuscript, 2008].

7. Linking the Sedimentological Record and Transport Processes

Through the whole studied sequence a strong correlation of the grain size record with the physical stratigraphy can be observed (see Figure 7). The loess layers, which are associated with cold climate conditions, are dominated by the deposition of coarse-silt aeolian dust. In contrast, finer airborne material has been deposited within the pedocomplexes that represent temperate interstadial or interglacial environments. The Uf−m ratio reveals this pattern in the sedimentological record for the silt-sized fractions. Coarse and medium silt deposition dominates within the loess. For instance the OIS 6 and the OIS 4 loess is marked by dominantly coarse silt input. Fine silt was preferentially deposited during periods correlating with OIS 5, OIS 3, and the Holocene pedocomplex.

Figure 7.

Grain size variations, defined in Figure 6, are plotted on our revised chronostratigraphic scheme of the Remisowka loess sequence. A strong correlation of the grain size record with the physical stratigraphy is observed. Loess layers, associated with cold climate conditions, are dominated by the deposition of coarse-silt aeolian dust. In contrast, finer airborne material has been deposited within the pedocomplexes that represent temperate interstadial or interglacial environments. The Twin Peak ratio shows recurrent long-term trends within the grain size record on a glacial-interglacial scale that correspond with the physical stratigraphy and depict a clear climatically triggered signal. The Uf−m ratio reveals this pattern in the sedimentological record for the silt-sized fractions. As the cyclonic activity, through interaction with the seasonal shift of the Asiatic polar front, mainly affects the dust transport in the region, we hypothesize that the observed variations in the dust record are generated by the migration of the polar front on a glacial-interglacial scale (see Figure 8), and the long-term effects of its seasonal duration and permanency on its winter position.

This general pattern of predominant coarse or fine silt deposition (expressed in every single PSD, as well as in the entire record) reflects the strong sorting behavior of the aeolian system on the sedimentological record. Stronger winds during glacials, favored owing to the steeper meridional contrast between southern tropical air and arctic air masses as well as the general glacial environment, carry coarser particles. Weaker winds, predominant during interglacials, are able to keep finer particles in suspension or transport coarser particles only over short distances.

The analysis of the Twin Peak ratio widens the perspective on the dynamics of Late Pleistocene dust deposition in SE Kazakhstan, respectively central Asia, showing recurrent long-term trends within the dust record on a glacial-interglacial scale, corresponding with the physical stratigraphy and depicting a clear climatically triggered signal. From the maximum of the penultimate glaciation (OIS 6) to the maximum of the superposed interglacial complex (OIS 5e), a gradual shift from coarser to finer silt deposition takes place. The same recurring long-term trend can be observed from the transition of the OIS 4 maximum to the overlying Holocene pedocomplex. At the transition from a pedocomplex to an overlying loess layer (e.g., OIS 5e to OIS 4) the general tendency turns in the opposite direction and a reverse gradual shift from finer to coarser silt deposition occurs. Particularly noticeable is the fact that these long-term trends are registered on an interglacial-glacial scale; the late stage 5 interstadials OIS 5c and OIS 5a, as well as the OIS 3 interstadial, slightly modify, respectively accentuate the general gradual shift, but they do not reverse or interrupt it (see Figure 7).

The observed variations in the bimodal silt distribution point out that Pleistocene glacial-interglacial climate dynamics, recorded in the studied loess sequence, are not only characterized by the already known coarse dust/fine dust deposition pattern, related to wind strength and thus to climate. Moreover it becomes apparent that a continuous force triggered the aeolian dust sedimentation on a glacial-interglacial scale. The recurrent long-term trends in the dust transport record represent lasting changes from glacials to interglacials and vice versa, affected by decreasing or increasing aeolian dust transport activities.

Although the grain-size record cannot provide information about the source of the aeolian material, we infer from the modern synoptical climatic patterns in central Asia and from studies on the paleoenvironmental development of the region [Walter, 1974] that the dynamics in the dust transport record are a consequence of changes in the regional or local aeolian activities. As the cyclonic activity through interaction with the seasonal shift of the Asiatic polar front mainly affects the dust transport in the region, we hypothesize that the observed variations in the dust record are generated by the migration of the polar front on a glacial-interglacial scale and the long-term effects of it seasonal duration and permanency on its winter position. It should be noted that physical-chemical properties of sediments are often controlled by annual or even subannual environmental variations, even though they do not reveal annual, or even decadal or centennial, layering. This however means that one can recognize long-term mean seasonal signals in loess sedimentary archives, despite the absence of annual temporal resolution.

8. Impact of the Asiatic Polar Front on the Aeolian Dust Dynamics in Central Asia

The polar front is a discontinuous border zone that generally separates the moister tropical air to the south from the drier polar air to the north [Harman, 1991; Shriner and Street, 1998]. The temperature and pressure differences between these two air masses are settled by the development of the near-surface midlatitude cyclones and the development of troughs in the polar front jet stream. In the Northern Hemisphere the polar front shows a seasonal movement southward in the winter and northward in summer, while in the winter months the frequency and strength of cyclonic storms is increased owing to the higher temperature and humidity gradient between the colder polar air and the warmer tropical air [Harman, 1991].

The general nature of the polar front is well understood [cf. Douglas, 1939; Harman, 1991; Burnett, 1993]. Several studies have illustrated the (paleo)climatic significance of the varying position and strength of the front (also referred to as the polar vortex) on annual to geological timescales [COHMAP Members, 1988; Burnett, 1993; Kirby et al., 2002]. On the basis of Atlantic deep-sea sediment cores, Ruddiman and McIntyre [1981] exemplified the role of the polar front shift during the climatic fluctuations of the last deglaciation for the North Atlantic, with a recurrent northward retreat during warming intervals and southward advances in course of cooling intervals, e.g., the Younger Dryas period. Other studies outlined similar patterns for northeastern North America [Kirby et al., 2002] or regions proximal to the North Atlantic in western Europe [Diefendorf et al., 2006], as well as for regions at the eastern fringe of the Eurasian landmass close to the Pacific [Schöne et al., 2004; Ono and Irino, 2004].

For interior Eurasia only a few observations have been previously noted that refer to the paleoclimatic impact of the migration and strengthening of the Asiatic polar front [e.g., Porter and An, 1995]. However, considering the modern synoptical atmospheric circulation and the pronounced seasonal-scale dynamics of the Asiatic polar front in interior Eurasia, we are led to the inference that the polar front showed a similar meridional migration and strengthening pattern on a glacial-interglacial and even shorter timescale. As the cyclonic activities along the polar front are one of the major driving forces on the aeolian dust transport in central Asia today [e.g., Franz, 1973; Lydolph, 1977; Liu et al., 2004] we hypothesize that the observed long-term trends in the dust transport record thereby reflect the long-term migration, seasonal duration and permanency of the Asiatic polar front associated with the climate shifts of the Quaternary Period.

In contrast to NE North America or the North Atlantic, the migration of the Asiatic polar front on the Eurasian continent experiences a major obstruction by the high mountains of central Asia that form a distinctive barrier blocking the meridional air mass transport and thereby restricting the polar fronts southern advance as well as favoring the development of the high-level planetary frontal zone (see Figure 1 and Figure 8). Hence, we understand the glacial-interglacial climate impact of the polar front not only expressed in the dynamics of an N-S migration but moreover in the concentrated intensity at its defined southernmost position. In fact, it seems likely that the paleoclimatic effects of the polar front movement are better represented by the long-term impacts of its changing seasonality from colder to warmer intervals and vice versa.

Figure 8.

The position of the Asiatic polar front in (a) winter and (b) summer, standing representatively for its migration and seasonal duration on a glacial-interglacial scale. The migration of the Asiatic polar front on the Eurasian continent experiences a major obstruction by the high mountains of central Asia that form a distinctive barrier, blockading the meridional air mass transport and thereby restricting the polar front's southern advance as well as favoring the development of the high-level planetary frontal zone. The steeper meridional contrast between southern tropical air and polar air masses and pressures, plus the lack of moisture during cold intervals and glacials, leads to a longer seasonal duration, permanency, and strength of the polar front on its southern winter position, involving increased and stronger cyclonic activity and aeolian transport. During warm intervals the seasonal duration and permanency of the polar front at the winter position are diminished, leading to more moderate cyclonic and aeolian activity, weaker winds in the region, and a longer retreat or even rare development at its northern summer position.

The steeper meridional contrast between southern tropical air and polar air masses and pressures, plus the lack of moisture during cold intervals and glacials lead to a longer seasonal duration, permanency and strength of the polar front on its southern winter position (seasonal domination), involving increased and stronger cyclonic activity and aeolian transport. During warm intervals, with a synoptical climate configuration similar to today, the seasonal duration and permanency of the polar front at the winter position is diminished, leading to more moderate cyclonic and aeolian activity, weaker winds in the region, and a longer retreat or even rare development at its northern summer position (Figure 8). Finally, we suppose that these lasting effects of changes and dynamics in seasonality are paleoclimatically recorded on a longer-scale in our observed particle size record. And, as the polar front activity is intimately connected with the development and position of the planetary frontal zone (HPFZ), the data give insight to long-term aeolian dust dynamics and climate variability of interior Eurasia that are linked with interhemispheric climates.

Even though our proposed model is mainly based on modern synoptical climatic patterns as well as on analogies with paleoclimatic observations made for the North Atlantic and the east Pacific, we see our hypothesis strongly corroborated by observations and models based on the modern Asian dust activity on annual and decadal scales by Yang et al. [2007] and Hara et al. [2006]. Also our ongoing studies and investigations on other loess sequences in SE Europe, central Asia and China show similar patterns on a wider spatial scale.

9. Conclusion

Aeolian dust dynamics recorded in the SE Kazakhstan loess sequence at Remisowka carry a highly resolved signal of glacial-interglacial paleoclimatic change. The application of highly resolved grain size investigations in combination with different geochronological techniques and chronostratigraphic tools, including luminescence and radiocarbon methods, plus amino acid geochronology, open up new vistas to investigate the aeolian dust record of terrestrial sediments. As those sediments are widespread on the continents and offer multiproxy investigations on regional and interhemispheric scales, we anticipate that the implementation of new methods will lead to a renaissance in the study of terrestrial climate archives.

On the basis of the results of the grain-size record, the generally accepted scheme of an interglacial preponderance of fine aeolian dust deposition and prevailing coarse dust sedimentation within glacials is confirmed. By studying the highly resolved grain size record of the SE Kazakhstan loess, with the emphasis on the bimodal distribution of the silt fraction, recurrent long-term trends of climate forcing on the aeolian dust dynamics through the Last Glacial Cycle in Eurasia are observed.

Modern synoptical atmospheric circulation patterns as well as studies in the North Atlantic region support our hypothesis that the observed dynamics reflect the long-term migration, seasonal duration and permanency of the Asiatic polar front on a glacial-interglacial scale.

Previously published models, which focused on the glacial-interglacial reciprocity of the zonal Westerlies and Asiatic high in central Asia, were overly simplified and should be modified to include the greater influence of the Asiatic polar front. In future paleoenvironmental studies of Eurasia, the influence of the Asiatic polar front must be considered as a significant factor forcing aeolian dust dynamics and associated environmental processes. As the polar front activity is intimately connected with the development and position of the high-level planetary frontal zone the data give insight to long-term aeolian dust dynamics and climate variability of interior Eurasia that are linked with interhemispheric climates.

Acknowledgments

This paper is dedicated to the memory of Andrei E. Dodonov. This work is part of an ongoing research program on aeolian dust dynamics and the reconstruction of past atmospheric circulations along the Eurasian loess belt funded by the German Federal Environmental Foundation (DBU). B.M. acknowledges receipt of additional valuable funding support from the German Academic Exchange Service (DAAD) and the von Humboldt-Ritter-Penck foundation for fieldworks from 2004 to 2007. Invaluable guidance and logistic help in Kazakhstan were provided by Aldar Gorbunov, Volodja Uvarov, and Alissa Uvarov. Special thanks to Paul Sanford for critical assistance in the USF Amino Acid Geochronology Laboratory. Thanks are extended to Frank Köhler and the Museum für Naturkunde, Humboldt University of Berlin, and Stefan Meng, who provided help and material to identify the gastropod shells for amino acid analyses. We thank Mladjen Jovanovic and Tivadar Gaudenyi for the legwork in the field in 2006. Jan-Berend Stuut and Maarten A. Prins are acknowledged in particular for their enthusiasm in bringing the aeolian dust community together several times and initiating this theme.

Ancillary